Offshore admins — £400 a month, and what it buys

The rate is set and stays set. £400 a month per head, as recorded in the desktop guide. Nothing on this page proposes changing it — the question it answers is what that £400 buys, and how the number of packs an admin gets through decides whether it is cheap or expensive.

The whole economics is throughput

A fixed salary means cost per pack is decided entirely by how many packs get finished. A full company is of work, so a month of hours has room for in theory — and nothing like that in practice once training, checking, waiting and everything else is taken out.

Two admins at ten packs a month cost more per pack than one at twenty. That is the whole argument for caring about throughput rather than headcount. It is also why the batch of ten matters: a whole company finished is measurable, ten unrelated tasks across nine companies is not.

The standard time is the control

This is what makes a fixed salary work. With no per-pack payment there is no automatic pressure to finish, so the pressure has to come from an agreed standard time per client — a number both sides know before the work starts, measured afterwards. Not a stopwatch; a benchmark.

And the numbers already line up. A batch of ten tasks averages — one working day, almost exactly. Ten a day was the right batch size before anybody worked out why.

The standard

a full client, start to finish

Range depending on condition — a clean company is quick, one on a strike-off with no Gateway is not. Hold to the average across a batch, never to the number on a single client.

What that means a month

packs, at 70% of a 147-hour month

147 productive hours is 7 a day over 21 days. The 70% allows for training, checking, rework and the days that go wrong. Assuming 100% is how a plan becomes fiction.

Hold them to touch time, not to the calendar. of tasks are waiting on somebody else — Companies House posting an auth code, HMRC deciding a VAT registration, a director finding 45 minutes. An admin cannot be late for HMRC.

Measure the hours they actually spend and whether the request went out on the day it should have. Measuring elapsed time instead punishes whoever was handed the companies with the most post in them, which is exactly the wrong signal.

How to run it

Quote the standard when the batch is handed over

"Ten tasks, about seven hours, back by end of day tomorrow." Both sides then know what good looks like, and there is nothing to argue about afterwards.

Compare against the batch, not the client

The slowest client here is more than twice the fastest. Judging a single pack against the average is unfair in both directions; judging ten is fair.

Log the actual time for the first fifty tasks

Every figure on this page rests on an estimate nobody has timed. Fifty tasks is enough to replace the estimate with a measurement, and then the standard is real.

When it slips, ask which half slipped

Over on touch time is a training or difficulty problem. On time but nothing finished is a waiting problem, and that is ours to fix by posting the slow requests earlier.

Never let the standard beat first-time-right

A pack finished inside the standard that bounces back from Submitted has cost more than one that ran over. Speed is the second measure, always.

Against the UK bookkeeper

Offshore, £400 a month

per pack at a realistic 20 a month

Fixed cost, variable output. Gets cheaper per pack the more that gets done, and more expensive if throughput drops.

UK bookkeeper, agreed rate

£125per pack, fixed

Variable cost, fixed price. You pay the same whether she does five or fifty, and she carries the risk of a slow one.

They are not really competing. One is a fixed cost you want to saturate; the other is a variable cost you pay only when work lands. The offshore team is the floor and the bookkeeper is the surge. Send the offshore team a steady queue and use the UK rate for whatever will not wait.

"More packs" is not an incentive on a fixed salary

This is worth being straight about. If an admin is on £400 a month whatever happens, then proving themselves and being rewarded with more packs means doing more work for exactly the same money. Nobody speeds up under that arrangement — the rational move is to do enough and no more. The intention is right; the mechanism does not carry it.

What actually makes it work. Keep £400 as the base and add one of these:

A bonus above a threshold. £400 covers the first packs a month; anything beyond is paid per pack. Costs nothing until the throughput is real.
A rate step for proven quality. Pass a defined number of packs at first-time-right and the base goes up. That is what "as good as the UK bookkeeper" should buy.
More packs meaning more heads. If volume grows, grow the team rather than the individual workload — that keeps the per-head rate honest and the throughput real.

What "as good as the bookkeeper" has to mean

If more work is the reward for quality, quality needs a definition that is not a matter of opinion. The dispatch portal already records everything needed.

Measure	Where it comes from	Why it is the right one
First-time-right rate	Tasks reaching Done without being sent back from Submitted	The only measure that separates fast from careless. A pack that comes back is worse than one that took longer.
Packs finished per month	Count of clients with every task confirmed	Whole companies, not tasks. A part-finished company has delivered nothing.
Priority-1 handled correctly	The critical board in dispatch	Destructive-if-ignored work is where a mistake actually costs something.
Evidence quality	What is attached at Submitted	Supreme has to check every one. Poor evidence moves the cost onto the checker rather than removing it.

Measure first-time-right before throughput, not after. Rewarding volume alone produces volume — and the checking burden lands on Supreme, where it is far more expensive than the time it saved. Nothing counts until it reaches Done.

What the piece rates in the system are for

The per-task rates in _data.js total across the estate, about a company. On a fixed salary those are not wages — they are an internal costing model, useful for comparing one runbook to another and for pricing a batch if it ever goes to somebody outside the team. Keep them consistent for that reason, but do not confuse them with what anybody is paid.

The correction made today matters for exactly that reason — a task that costed at £1 an hour against a £6 median made every comparison built on it wrong.